why do workers join trade unions?
TRANSCRIPT
Tanzanian Economic Review, Vol. 9 No. 2, December 2019: 119–140
©Department of Economics, University of Dar es Salaam, 2019
Why Do Workers Join Trade Unions?
Damiano Kulundu Manda & Germano Mwabu*
Abstract
This paper uses data from manufacturing firms in Kenya to analyse the
determinants of trade unionism. Matched worker-firm data from a recent Kenya
manufacturing enterprise survey is used to estimate a reduced-form probit model of
union membership. The results show that the likelihood of being a union member
decreases with schooling; but is higher for full-time, married, and city employees. We
conclude that Kenyan workers become union members to protect job tenure and to
improve working conditions.
Keywords: trade union membership; union density; manufacturing; Kenya.
JEL Classification: J32, J51, J52
1. Introduction
The behaviour of the labour market in many countries is to some extent
influenced by trade unions and their role as representatives of the workforce in
collective bargaining with employers or with government. The ability of trade
unions to influence the conditions of pay and work in the labour market largely
depends on the degree of unionisation of the workforce; that is, the fraction of the
workforce in trade unions (Checchi & Corno, 1998). However, the effectiveness of
trade unions in influencing conditions of pay and work depends on the degree of
membership participation in trade union activities (Anyango et al., 2013). Trade
unions thus derive their legitimisation to represent the interests of the workforce
from a high degree of union density and union members’ participation in union
activities. A large degree of unionisation and participation enhances the
bargaining power of a trade union, and enables it to negotiate for higher pay rises
and to improve working conditions (Checchi & Corno, 1998).
The main objective of trade unions is to raise and protect the welfare of their
members (Manda et al., 2005; Anyango et al., 2013). This objective can be
achieved through various means, such as improved pay and work conditions,
support in the event of a problem at a workplace, job security, information about
employment opportunities, legal advice, industrial benefits (e.g., pensions,
medical insurance, housing allowances, etc.), training, education, financial aid,
and social network. Thus, apart from joining trade unions to push for higher wage
levels, there are non-wage benefits that attract individuals into labour unions.
*School of Economics, University of Nairobi: [email protected]
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
120
In Kenya, labour unions are usually found in manufacturing, trade, transport,
large-scale agriculture, public and teaching sectors. Union membership is
voluntary: it is estimated that about one-third of all Kenyan workers in the
formal manufacturing sector are unionised. This is a relatively sizeable fraction of
the labour force, given that payment of union dues is a prerequisite for becoming
a union member. Data from Kenyan manufacturing firms for the period 1993-
1995 show that about 40% of production workers belonged to trade unions, which
is well over the national average of about 30% of employees then. The same data
shows that the proportion of workers in trade unions varies from one region to
another. For instance, 44% of the workers in manufacturing firms in Nairobi are
union members compared with about 30% in Mombasa and Nukuru, and 25% in
Eldoret (Manda et al., 2005). It is unclear why there are these variations in union
membership. Further, it is puzzling that a larger fraction of Kenyan workers are
not union members despite the large wage and non-wage benefits associated with
being a union member. It is therefore important to understand why some workers
join trade unions and others do not. This study analyses the determinants of
union membership to shed light on this issue.
The remainder of the paper is structured as follows: section 2 presents theories of
trade unions, and section 3 presents previous empirical studies. The model used,
data and discussion of variables used are in section 4, while section 5 discusses
the results. Finally, section 6 deliberates the conclusion.
2. Theories of Trade Unions
There are several theories that have been developed on trade unionism. These
include the political revolutionary theory of labour movement, theory of industrial
democracy, theory of union control of industry, environment theory, theory of
man vs. machine, and the theory of the ‘scarcity consciousness’ of manual
workers. We summarize each of these theories below.
The political revolutionary theory of labour movement was proposed by Marx and
Engels, and is based on Adam Smith’s theory of labour value (see Smith, 1776). The
theory sees trade unions as simply a class struggle between proletarian workers
and capitalist businesspeople with a short-run purpose of eliminating competition
within labour, and ultimately leading to the overthrow of capitalist businesspeople.
According to Marx, trade unions are instruments to overthrow capitalism.
Next is the theory of industrial democracy by Webb and Webb (1897). According
to the theory, trade unionism is an extension of democracy from political sphere
to industrial sphere. Like the political revolutionary theory, the theory agrees
that trade unionism is a class struggle but looks at trade unions as providing
means to workers to overcome managerial dictatorship and express their voice in
the determination of the conditions under which they work. In other words, the
theory considers trade unions not as an instrument to overthrow capitalism, but
as a means of equalizing the bargaining power of labour and capital. Webb and
Webb considered collective bargaining as a process that strengthens labour.
Why Do Workers Join Trade Unions?
Tanzanian Economic Review, Volume 9, Number 2, 2019
121
Third is the theory of union control of industry propounded by Cole (1913). The
views of the theory are somewhere in between the political revolutionary theory
and the theory of industrial democracy. While the theory agrees that unionism is
a class struggle, the ultimate aim of the struggle is the control of industry by
labour and not revolution as predicted by Marx.
Fourth is the environment theory propounded by Commons (1913). The theory is
sceptical of generalisations and agrees that collective bargaining is an instrument
of class struggle. However, it points out that, ultimately, there will be partnership
between employers and employees.
The theory of man vs. machines—also sometimes referred to as the ‘rebellion
theory’—was advocated by Tannenbaum (1921). According to this theory, the use
of machines leads to the exploitation of workers; therefore, machines are the
cause and trade unions are the outcome. According to the theory, a trade union is
formed in reaction to the alienation and loss of community in an individualistic
and unfeeling society. The union gives a worker a fellowship and a value system
that s/he shares with other workers. Institutionally, a trade union movement is
an unconscious effort to harness and reorganise around the cohesive identity that
men working together always achieve.
Finally, is the theory of the ‘scarcity consciousness’ of manual workers by Perlman
(1928). The theory rejects the idea of class consciousness as an explanation for the
origin of a trade union movement, and substitutes it with what is known as job
consciousness. According to the theory, “… working people in reality have an urge
towards collective control of their employment opportunities, but hardly towards
control of industry.” The theory further proposes that a theory of the labour
movement should include a theory of the psychology of the labouring man. It is
when manual workers become aware of a scarcity of opportunity that they band
together into unions for the purpose of protecting their jobs and distributing
employment opportunities among themselves equitably, and to subordinate the
interests of the individual to the whole labour organism.
Generally, the theories of trade unions point to the reasons as to why workers
may want to join trade unions. Whereas researchers have devoted time and
effort to study ‘why employees choose to join a union,’ findings of their studies
do not report a common list of reasons. Nonetheless, there is a general
agreement among labour experts that certain issues are likely to lead to an
organizing drive by workers. First, from an economic point of view, workers act
rationally and therefore they join a union with the aim of maximizing their
benefits through union membership. The excess of benefits over costs justifies
workers’ joining a trade union. Some of the reasons why workers join a union
include: job security, wages and benefits, working conditions, fair and just
supervision, powerlessness, and the need to belong. Our analysis in this paper
looks at some of these reasons. The next section looks at previous empirical
studies.
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
122
3. Previous Empirical Studies
There is a large and diverse literature on the determinants of union membership
in developed and developing countries, Kenya included. We review only a few to
highlight common analytical methods and findings.
Studies based on time series data (e.g., Komsi, 2010; Schnabel & Wagner, 2005;
2007; Borland & Ouliaris, 1994) find that business cycle factors and structural
developments are important macro-determinants of union membership. Studies
based on cross-section and panel data (e.g., Windolf & Haas, 1989; Fitzenberger et
al., 2009; Kollmeyer, 2007) show that trade unionism is determined by personal,
occupational and firm characteristics; earnings; attitudes; social variables; and
institutional factors such as unemployment insurance benefits.
Most studies done on trade unions in developing countries, Kenya included, are
based on cross-sectional data (see, e.g., Rupayan, 2008; Guataqui et al., 2011;
Anyango et al., 2013; House & Rempel, 1976; Manda, 1997; Manda et al., 2005;
Schultz & Mwabu, 1998). Due to data limitations, some of the studies focus on
certain groups of factors affecting union membership. For instance, Guataqui et
al., 2011) focuses on structural determinants of trade union membership in
Colombia; while the study by Rupayan (2008) looks at political factors on union
membership in addition to individual and firm level factors. In Kenya, some
authors (e.g., House & Rempel, 1976) show that, for certain categories of workers,
trade union membership is negatively associated with earnings, or has no effect
on earnings. The study by House and Rempel (1976) does not, however, offer an
explanation as to why workers join trade unions if union membership is not
associated with an increase in earnings. Manda et al.’s (2005) study on Kenya
shows that the negative relationship between union membership and earnings
could be due to flaws in the methodology used to analyse union membership.
They argue that endogeneity of union membership in the earnings equation is the
likely reason for the negative relationship reported in the literature (ibid.). The
Kenyan studies, however, do not provide a comprehensive analysis of union
memberships. This study aims to fill this gap by offering a rigorous analysis of
why some individuals are union members while others are not.
We use worker-firm merged data on Kenyan manufacturing sector to analyse factors
affecting union membership. We look at factors commonly identified in previous
studies with a focus on age, education, occupation, working status; and workplace
characteristics such as union density, firm size, location, and type of industry.
4. A Model of Trade Union Membership
4.1 Introduction
Assuming that the objective of a trade union is to maximise the well-being of its
members, we ask two related questions: First, what do trade unions deliver to
workers to achieve this objective; and what means do they use to do so? Second,
what do workers expect to get from trade union membership, and what factors
affect the perceived benefits of membership?
Why Do Workers Join Trade Unions?
Tanzanian Economic Review, Volume 9, Number 2, 2019
123
The first question has to do with the supply-side of trade union membership, and
deals with the benefits offered to members and the “things that trade unions do” to
facilitate membership. As in other market settings, the benefits of membership are
offered to workers at some positive price. The second question relates to the
demand-side: that is, workers’ propensity to join trade unions to acquire the
associated benefits. This propensity is determined by the sacrifice a worker must
make to join, conditional on the work environment and socio-economic status.
Assuming that the activities of a trade union are not subsidised, the price that
workers pay to join a trade union (say, annual membership fee) cannot be less than
the cost of the benefits provided to them by the union. Under the no-subsidy
assumption, the level of membership fees charged by a trade union depends on the
cost of providing membership services. Thus, since production and cost structures of a
trade union affect membership fees, they also affect membership decisions. It should
be noted that a trade union is both a firm, and a consumer. As a provider of services
that improve the working conditions of workers, a trade union is a firm; whereas, as
an end-user of its own services, it is a consumer.
We focus on trade union as a collective consumer and explore how an individual’s probability of joining this collective firm is affected by its own characteristics, and
by the characteristics of the employer. We make a simplifying assumption that
union leaders and union workers have the same preferences over the services
provided by a union so that the behaviour of both can be described by the common
preferences assumption. The common preferences assumption is rather innocuous
given that the benefits from union membership are a form of a public good. For
example, a pay increase negotiated by a union may be extended to all workers
irrespective of their union status (Schultz & Mwabu, 1998). Moreover, union
protection from unfair dismissal is available to all members. Because of the public
good nature of union benefits, all paid-up members may not make the same non-
monetary effort towards the strengthening of the union power.
In the literature, the free rider problem is usually identified with non-unionised
workers who benefit from union spill-over benefits (Booth, 1986). However, as we
have done here, the free rider problem can also be identified with union members
who do not work to increase union power because they know others are working
to do so. Nonetheless, Cregan and Johnson (1990) show that pro-union norms and
peer pressure at workplaces in favour of unions may be effective checks against
shirking and free riding. Thus, despite the public goods’ characteristics of benefits
from union membership, we continue to maintain a common preference
assumption, in which the identity of a union member is unimportant in achieving
the overall objective of the union. However, the free rider hypothesis in union
membership can be tested directly (Booth, 1986).
In the framework developed below, a worker chooses to join a union if the
perceived net benefit of union membership is greater than the benefit of not
belonging to a union. We analyse workers’ trade union membership decisions in
Kenyan manufacturing using a simple qualitative choice model of union status.
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
124
4.2 Framework of Analysis
Following standard practice in the formulation of qualitative choice models (see,
e.g., Schmidt & Strauss, 1976; Booth, 1986), the net benefit (utility) from trade
union membership for worker i can be expressed as:
𝑉𝑖 = 𝑉(𝑃1, 𝑃2, … 𝑃𝑛 , 𝑌, 𝑆, 𝑍) = 𝑀𝑎𝑥𝑈𝑖 = 𝑈(𝑋1, 𝑋2, … 𝑋𝑛) subject to 𝛴𝑃𝑖𝑋𝑖 = 𝑌 (1)
Where, suppressing the subscript for the worker:
V(.) is the indirect utility function for a worker;
U(.) is the corresponding direct utility function;
P1… Pn-1 are prices of goods and services supplied by the union, with Pn
being a price index for non-union goods, which are assumed to be a
composite commodity;
X1 … Xn-1 are goods and services supplied by the union, with Xn being a
composite commodity that is supplied outside of the trade union;
Y is the worker’s resource endowment, which determines maximum welfare
when it is efficiently allocated; while
S and Z denote, respectively, a worker’s social and job characteristics.
Expression (1) illuminates several issues that are unclear in the earnings-union
literature. First, the direct utility part of the expression shows the range of goods
and services (the Xs) supplied to the worker by the union. These goods and
services (Deery & Cieri, 1991; Waddington & Whitson, 1997) include:
(a) Improved pay and work conditions;
(b) Support by the union if there is a problem at the place of work;
(c) Job security;
(d) Quality employment opportunities;
(e) Legal advice;
(f) Industrial benefits such as allowances, pensions, medical insurance, and
housing;
(g) Training and education;
(h) Financial services;
(i) Professional support;
(j) Companionship; and
(k) Opportunity to advance one’s social goals.
The indirect utility part in equation 1 shows the maximum welfare level that a
worker can attain given the prices of ‘union goods’ (i.e., goods (a) through (k)
above), his/her endowment Y, and the prevailing social and industrial
environment as captured by S and Z. The indirect utility part is especially useful
from an empirical viewpoint because it contains observable information about a
union’s resource endowment, and the vector of prices of the goods that a union
can provide, including unobservable goods such as job security. Solving equation
1 one obtains the following demand function:
𝑋𝑖𝑗 = 𝑓(𝑃𝑗 , 𝑆𝑖 , 𝑍𝑖, 𝑌) (2)
Why Do Workers Join Trade Unions?
Tanzanian Economic Review, Volume 9, Number 2, 2019
125
Where 𝑋𝑖𝑗 is the quantity of ‘union good’ j demanded by worker i. Conditional
on being a union member, monetary prices for these goods, the 𝑃𝑗s, are zero.
For example, union members receive wage increases negotiated by union leaders
free of charge. From expressions 1 and 2, workers may join a trade union for a
variety of reasons. For example, they can sacrifice a higher wage to receive a low
wage in unionised firms for reasons of job security and job satisfaction; to be in
company of people with ethnic or social values that they share; or to receive social
services such as pensions and medical care that are not available in non-
unionised firms. Thus, the sign of the effect of union membership on quantities of
goods and services shown in equation 2 is an empirical matter.
It now remains to show the conditions under which a worker will join a trade
union.
Let 𝑉𝐽and 𝑉𝑁 be the indirect utilities of joining and not joining a trade union,
respectively. Following Booth (1986) and suppressing subscripts for individuals, a
worker i will join a union if:
𝑉𝐽 − 𝑉𝑁 > 0 for 𝑖 = 1, … , 𝑛 (3)
with the deterministic and stochastic components of the indirect utilities
taking, respectively, the following forms:
𝑉𝐽 = 𝑋𝐽𝛽 + 𝜉𝐽, and similarly 𝑉𝐽 = 𝑋𝑁𝛽 + 𝜉𝑁 (4)
Given equation 3, the probability 𝑃𝐽 of worker i joining a union can be expressed as:
𝑃𝐽 = prob {(𝑉𝐽 + 𝜉𝐽) − 𝑉𝑁 + 𝜉𝑁) > 0} (5)
Which, by rearrangement becomes:
𝑃𝐽 = prob (𝑋𝐽𝛽 − 𝑋𝑁𝛽 > 𝜉𝑁 − 𝜉𝐽) (6)
Assuming the error terms in equation 6 are independently and identically
Weibull distributed, the cumulative joint density function for the errors—i.e.,
the unionisation probability—can be expressed in conditional logit form
(Rupayan, 2008; Kollmeyer, 2007). If the error terms follow a normal
distribution, a normal cumulative distribution function is the appropriate
statistical description of unionisation probability, and a probit model happens.
We estimate a reduced form probit model to analyse determinants of union
membership in using Kenyan data. The next section describes the data and
the variables used for estimation.
4.1 Data and Variables
We use matched worker-firm data from the Kenyan Manufacturing Enterprise
Survey (KMES) carried out in 2000. The KMES was carried out by a team of
researchers from the University of Oxford, Gothenburg University, and the
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
126
University of Nairobi in conjunction with the Kenya Association of
Manufacturers. Much of the data gathered refer to the most recent financial
year of the firm, which means that such data refer to 1999. When firms were
asked about their current situation, their replies should be construed to refer to
the actual time of the interview. Also, most of the data on workers refer to the
year of data collection.
The KMES covered firms in four manufacturing sub-sectors (food, textile, wood,
and metal) in four major urban centres in Kenya (Nairobi, Mombasa, Nakuru,
and Eldoret). The sample is predominantly made of formal sector firms
(comprising of about 90% of the firms), and most (55%) are in Nairobi. The firms
were selected at random from a sample frame made up of different lists of firms
in the country. The formal firms were chosen at random from the master file of
registered firms from the Kenya National Bureau of Statistics. The sample frame
for informal firms was constructed by undertaking a primary listing of firms in
the four sectors, and then selecting the firms at random from the lists. The survey
collected information from firms and workers.
The labour market sections of the survey provide information on both firm and
individual characteristics. The information on firms was elicited from
representatives of each firm, and included data on the total number of firm
employees, total wage bills, profits, firm ownership, proportion of firm employees
in trade unions, proportion of workers that are permanent (fulltime), casual
employees, number of workers who quit the firm or changed jobs the previous
year, the value of plant and equipment, value added, among others. In all
probability, the data-set collected on individuals from firms is more accurate than
would have been the case had the information been collected from households.
The information on individuals was obtained by interviewing at most ten (10)
employees randomly chosen from a list of workers of each firm, among those
present on the premises on the day of the interview. This could be a potential
source of endogenous stratification. All employees in small firms were interviewed.
The data on individual workers contains information on education, job experience,
tenure, age, sex, marital status, hours of work, earnings, status of employment,
among others. In addition to these, the KMES provides information on the family
background of a worker, and whether an employee had changed jobs.
Following Booth (1986), Deery and Cieri (1991), Borland (1996), Waddington and
Whitson (1997), Rupayan (2008), Manda et al. (2005) and Guataqui et al. (2011),
we discuss variables that on a priori considerations are thought to affect union
membership. We classify the variables by worker and firm characteristics.
Individual characteristics that are likely to affect union membership as identified
in the literature are sex of the worker, working status (full-time or part-time),
firm-specific training, general training (typically obtained outside the firm),
family background and commitment, education, age or labour market experience,
tenure, occupation, and political ideology or loyalty.
Why Do Workers Join Trade Unions?
Tanzanian Economic Review, Volume 9, Number 2, 2019
127
4.1.1 Worker Characteristics
Starting with the sex of the individual, women are traditionally believed to be
less likely to unionise because of their discontinuous attachment to the labour
market due to family commitments or social norms that discourage them from
taking up paid work outside the home. Thus, relative to men, women are less
likely to unionise. We use a dummy variable to capture the effect of sex on union
membership. Also, in comparison with full-time workers, part-time employees are
less likely to unionise because they do not expect to be in a workforce for a long
time. For such workers, benefits from unionisation are short-lived and may not
materialise during periods of employment. We use a dummy variable to capture
the effect of working status on union membership.
Compared to workers with general training, employees with specific or on-the-job
training are more likely to protect their job tenure by joining trade unions
because they may be unable to use their skills outside of the firm. In Hirschman’s
(1970) terminology, workers with firm-specific skills are more likely to use voice
rather than threat of exit to protect their jobs. A dummy variable is used to
capture the effect of on-the-job training on union membership. Similarly, people
who have family commitments—such as caring for the young and the elderly—are
more prone to unionisation for reasons of job security. We use a dummy variable
for marital status to capture the effect of family commitments.
Education affects unionisation probabilities through several channels. Educated
workers may expect to achieve more in their careers through personal efforts
than through the bureaucratic machinery of a trade union and, thus, be unwilling
to unionise. Also, highly educated individuals sometimes identify more with
management than with a labour movement. Alternatively, workers with good
education may want to use a trade union to advance certain social causes because
they would, on average, be more effective in articulating their views. Thus, the
influence of education on unionisation probability is ambiguous to start with. We
measure education using total number of years of schooling for an individual, or
their level of education.
Depending on the benefits provided by unions, it may attract workers of different
age groups. For instance, if some unions provide health and old age benefits such
as medical and pension schemes, they will attract older workers. In contrast,
young workers may not be influenced by schemes such as choice of careers, which
they are likely to pursue independently of union availability. If, on the other
hand, working conditions and earnings are lower than what younger workers
expect, then they are likely to join a union if its union aim is to improve these
conditions. We capture the effect of this variables using years of work experience.
Occupation types influence the union status of workers. In some occupations in
Kenya (e.g., teaching, civil service, food, and allied services) workers have an
incentive to unionise because the skills they acquire for their jobs give them
power to extract rents from employers. Also, workers in occupations such as
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
128
production lines may have no restrictions to join a union compared to their
counterparts in management. We use a dummy variable for production workers
to capture the effect of occupation on union membership. We expect production
workers to have a higher probability of being union members.
Workers may also join trade unions if they believe that the unions will help
improve their working conditions. This is the case especially if the working
conditions at a workplace are undesirable. Workers in jobs with undesirable
characteristics (e.g., dangerous working environment, long hours of work) are
therefore likely to join trade unions or seek compensatory pay. In this paper, we
use weekly hours of work as a proxy for working conditions. Longer hours of work
without extra pay are associated with poor working conditions.
Some attitudes, social values and political ideologies are associated with greater
propensity to unionise. In Western countries (Deery & Cieri, 1991; Riley, 1997;
Kollmeyer, 2007), people with leftist political ideologies, those who participate in
political activities, and those who describe themselves as working class are more
likely to identify themselves with trade union causes and to voluntarily become
union members. In the Kenyan case, workers who identify themselves with
opposition political parties, or who are from ethnic groups that consider
themselves not treated well by the ruling party, are more likely to join trade
unions, taking them as alternative institutions for pursuing their social and
political ends. However, our data does not have information that can enable us to
control for the effect of this variable in our analysis.
Several theories of social psychology and social custom models suggest the inclusion
of social variables in the union membership model. The influence of key
individuals—such as parents and spouses—on the decision to be a union member
has been found to have mixed outcomes (Windoff & Haas, 1989; Goerke &
Pannenberg, 1998). Our data provides information on the education and occupation
of an employee’s father. We use a dummy variable on father’s education to control
for the effect of this variable on union membership.
As mentioned earlier, the free rider problem in the literature is usually identified
with non-unionised workers who benefit from union spill-over benefits (Booth,
1986). The free rider problem can be identified with workers who do not work to
increase union power because they know others are working to do so.
Nonetheless, pro-union norms and peer pressure at workplaces in favour of
unions may provide effective checks against shirking and free riding (Cregan &
Johnson, 1990). Also, in a median voter model in which workers have different
reservation wages and hence different optimal points in the trade-off between an
increased wage and a decreased probability of employment, Bulkley and Myles
(2001) argue that joining a union instead of free riding may be rational. This
would be the case if joining a union enables individuals to influence union
bargaining goals and thus their own employment probability. The free rider
hypothesis in union membership can be tested directly (Booth, 1986).
Why Do Workers Join Trade Unions?
Tanzanian Economic Review, Volume 9, Number 2, 2019
129
It is expected that trade unions would provide job security to their members.
Therefore, union members would enjoy greater job security than non-members so
that tenure would be higher for union than non-union workers. Therefore, job
security should act as an incentive to join a union.
4.1.2 Firm Characteristics
From the literature, firm level factors affecting union membership include the
following: state of labour demand in the industry, firm size, union density, type of
industry, organisational costs, union recruitment practices, location of firm
(industry) and/or workers, union-provided goods such as wages and allowances,
and collective bargaining (e.g., decentralised or centralised system). We provide
below a brief description and expected effects of some of these variables drawing
mainly on literature, following closely the format used by Booth (1986).
The state of labour demand in an industry has an important effect on the ability of
unions to negotiate wage increases and to secure other union-provided goods such
as fringe benefits for its members, factors that attract workers to unions. If labour
demand is wage inelastic, then trade unions have better chances of raising wages
and of increasing membership than if they face elastic demand schedules. The rate
of unemployment in an industry may be used as a proxy for the state of demand for
labour. However, we are not able to capture this variable due to data limitations.
Large firms are more likely to have unionised workers because the costs of
organising trade union activities in such firms tend to be low relative to costs in
small firms. Further, workers in large firms tend to be alienated from one another
by specialised job routines, and are more likely to be treated anonymously by
management in contrast to situations in small firms where face-to-face interaction
among workers is common. Workers in large firms are thus more likely to turn to
trade unions to fulfil socialisation needs, and to protect themselves from arbitrary
behaviour of management. To the extent that union organisational costs are lower
in large firms, union densities would also tend to be high in such firms, a situation
that is conducive to unionisation of new workers. Also, in such organisations, there
tends to exist higher peer pressure to conform to a social custom of union
membership (Riley, 1997; Schnabel & Wagner, 2005). We measure firm size by the
number of employees in a firm and its square. We expect positive but declining
impact of firm size on union membership, reflecting lower organising costs for
unions in large firms.
Recruitment practices of unions also affect union densities, with densities being
higher in firms where recruitment is face-to-face, especially in industries with
decentralised collective bargaining. However, decentralised bargaining has the
disadvantage that its negotiated wages are normally lower than wages secured
through a centralised process. Thus, an increase in unionisation due to
recruitment efforts at a firm level may be offset by inability to negotiate
sufficiently high wages. Thus, decentralised and centralised collective bargaining
affects unionisation through the wage rate in opposite directions. However, if
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
130
workers prefer social outcomes of a bargaining process over monetary outcomes,
decentralised bargaining may increase unionisation even though it is associated
with lower wages. The data that we have enables control for the effect using two
variables: centralised, and decentralised collective bargaining.
Due to the limitations of the survey data used in the analysis, only some of the
variables discussed above are included in the union participation equation that
we estimate. In particular, data on the state of labour demand, attitudes, social
values, and political loyalties are not available. The variables included in the
analysis are defined as shown in Table A1. They include, among others, the
dependent variable, i.e., union membership status (a dummy variable taking
value 1 if an individual is a union member, and zero otherwise).
5. Results
This section discusses results in two sub-sections. The first sub-section looks at
the descriptive statistics, while the second discusses the probit results.
5.1 Descriptive Results
Table 1 shows the means and standard deviations of the variables used in the
analysis. On average, the workers have about 10 years of education. The mean years
of experience for all workers is about 12 years; and 13 years for production workers.
Table 1: Means and Standard Deviation of Variables Used in the Analysis
Variables All Workers Production Workers
Mean Standard Deviation
Mean Standard Deviation
Weekly hours of work Age Years of education Experience On-the-job training General training Production workers Firm size (workers) Tenure (years) Married Located in Nairobi Male Food sector Wood sector Textile sector Metal sector Union density Central bargaining Decentralized bargaining Fulltime employees Formal sector Father has university education Foreign-owned farm Total Number of observations
46.4 34.8
9.8 12.2 0.11 0.45 0.61 162 8.6
0.75 0.55 0.81 0.29 0.19 0.26 0.26 33.2 0.06 0.44 0.76 0.92 0.03 0.11
1088
7.0 9.7 2.9 9.3
- - -
289 8.6
- - - - - - -
33.2 - - - - - -
46.7 35.3
8.9 13.1 0.14 0.34
- -
9.4 0.80 0.54 0.91 0.25 0.22 0.28 0.25 33.9 0.05 0.45 0.73 0.91 0.01 0.09 670
6.1 35.4
2.7 9.9
- - - -
8.1 - - - - - - -
33.9 - - - - - -
Why Do Workers Join Trade Unions?
Tanzanian Economic Review, Volume 9, Number 2, 2019
131
About 11% of the workers have on-the-job-training, while 45% have general training. Production workers constitute about 61% of the workers in our sample, and an average firm employs 162 workers. The average tenure for workers in the manufacturing sector is about 9 years. About 75% of the workers in our sample are married. Of the firms, 55% are in Nairobi, 25% in Mombasa, and the remaining in Nakuru and Eldoret. Most of the respondents (81%) are males, indicating that only 19% of the sample respondents are females; reflecting the
domination of the manufacturing sector by male workers. About 9% of the employees in our sample have changed employment. The proportion of employees working in foreign-owned firms is about 11%.
The average union density is about 33.2%. This indicates that about one-third of workers in the manufacturing sector are trade union members. This is not very different from the national estimates of about a third of the workers. As already noted, trade union membership is not compulsory in Kenya. The question is whether there is any distinct group of workers who join unions. We attempt to
answer this question using information in Table A2 in the Appendix.
Columns 1 and 2 in Table A2 show the means for union and non-union members based on the entire data-set. We can see from the table that both union and non-union workers are engaged on average for 46 hours in a week. Also, on average
the age of union workers is about 38 years compared with 34 for non-union workers. Non-union workers are more likely to be less educated as indicated by about 9 years of education for union workers, and 10 years of education for non-union workers. It is likely that older workers and less educated ones are likely to
join trade unions for reasons of job security. Job insecurity may arise from the fact that employers may prefer younger and highly educated individuals to older workers. It is also likely that many of the young workers, being highly educated, work in occupations such as management and administration, which do not allow one to join trade unions; or join firms where there are no trade unions.
Union workers have longer working experience than non-union workers. This probably confirms that older people are more likely to join trade unions than the young. Most of those with general training are non-union members. As shown in
Table A2, about 27% of the union workers and 50% of the non-union workers
have general training. On the other hand, 84% of production workers are union-workers, compared with 56% non-union production workers. This shows that production workers are a dominant group not only for union workers but also for non-union workers. Also, male workers form most of the union workers, although they also form most of the non-union workers. This, however, reflects the fact that male workers are the dominant workers in the manufacturing sector.
Another important result from Table A2 is that tenure is higher for union than
for non-union workers. This partly provides another possible reason why some workers may join a trade union. Workers possibly join a union for job security. Also, most union workers seem to work in larger firms than non-union workers. Lastly, union workers are in firms that are more unionized. This suggests that the union density of a firm may have some influence on union membership.
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
132
Table 2: Means for Selected Variables by Location and Union Status
Variables All Workers
Nairobi Other Towns*
Union Non-union Union Non-union
Weekly hours of Work
Age
Experience
Years of education
Firm size (workers)
Tenure (years)
Union density
Total observations
44.7
37.0
14.6
8.6
148.0
11.7
68.4
145
45.0
36.6
11.8
10.1
164.0
7.7
29.9
438
48.3
38.5
8.3
16.1
199.0
14.0
62.5
75
47.8
33.5
10.0
11.0
156.7
7.3
18.8
485
Production Workers
Nairobi Other Towns*
Union Non-union Union Non-union
Weekly hours of work
Age
Experience
Years of education
Firm size (workers)
Tenure (years)
Union density
Total observations
48.3
38.5
8.3
16.1
199.0
14.0
62.5
75
45.4
35.0
9.1
12.5
131.4
8.1
28.0
229
48.6
40.6
7.6
18.3
190.4
16.2
61.7
59
48.4
33.7
9.1
11.3
123.7
7.7
14.0
244
Note: * The other towns are Mombasa, Nakuru and Eldoret
Table 2 shows that the average age and experience for union workers is generally
higher than for non-union workers. The average number of years of education is
generally higher for non-union workers than for union workers. Union density is
highest among firms with union members in all locations, and firms with union
members are also generally larger than those without union members. Also, tenure
is higher among union than non-union members, with years of service for union
workers in the current firms being highest in Mombasa, Nakuru, and Eldoret.
5.2 Probit Results
Table 3 presents the probit maximum likelihood estimates of the effect of both
individual and industrial characteristics on workers’ participation in trade unions.
The results are marginal changes in the probability of being in a trade union
following an infinitesimal change (assumed to be a unit change) in factors that are
theoretically important in influencing that probability. As the table shows, the t–
ratios associated with the marginal effects show that most of the variables included
in the model influence union membership. Since production workers comprise 61%
of all workers in the manufacturing sector, we estimate a separate membership
probit equation for this category of workers (last two columns of Table 3).
As Table 3 shows, the marginal effect associated with years of education is negative.
This means that workers with higher education are less likely to be trade union
members as shown by both sets of results (all workers, and production workers).
Why Do Workers Join Trade Unions?
Tanzanian Economic Review, Volume 9, Number 2, 2019
133
Table 3: Probit Analysis of Trade Union Membership
Variables All Workers Production Workers
Marginal
Effects
T-value Marginal
Effects
T-value
Years of education
Experience
Experience squared
On-the-job training
General training
Male
Full-time employment
Union density
Father has university
education
Located in Nairobi
Food sector
Wood sector
Textile sector
Weekly hours of work
Production workers
Foreign firm
Firm size
Firm size squared
Tenure
Non-members benefit
Formal sector
Central bargaining
Decentralised bargaining
Married
Number of observations
Log likelihood
Pseudo R-squared
-0.01597
-0.01088
0.00007
-0.0569
-0.09529
0.02730
0.08113
0.00282
0.28263
0.0354
0.1304
0.00703
0.04590
0.00005
0.1186
-0.0347
0.0001
-0.0000
0.0067
0.09037
0.0635
0.0962
0.0436
0.0746
948
-297
0.40
-3.57
-3.05
1.14
-2.27
-3.77
0.87
2.93
7.30
2.71
1.64
3.76
0.20
1.39
0.03
5.42
-1.16
0.83
-0.38
2.97
3.43
2.13
1.61
1.29
2.81
-0.0184
-0.0147
0.0001
-0.0925
-0.1022
-0.0063
0.1524
0.0046
0.1384
0.0853
0.1905
0.0332
0.0749
0.0059
-
-0.0582
0.0001
-0.0000
0.0105
0.1299
0.0697
0.1733
0.0154
0.1183
616
-216.75
0.41
-2.51
-2.51
0.94
-2.16
-2.58
-0.09
3.15
6.92
0.50
2.34
3.22
0.57
1.4
1.72
-
-1.10
0.42
-0.73
2.92
2.99
1.26
1.68
0.27
2.47
An extra year of schooling is associated with a 1.59% increase in the
probability of union membership for all workers, and with 1.84% increase
among production workers. Another factor driving unionization is full-time
employment. The probability of joining a union for all workers is 8.1% higher
among full-time workers (relative to part-time workers), and 15.2% higher for
production workers.
The marginal effect of experience is negative, and that of experience squared
is positive, an indication that experience has a U-shaped relationship with the
probability of being a union worker. The explanation for this could be that
young workers who possibly have higher education are likely to occupy job
levels that put limits to joining trade unions, or jobs that provide more
benefits than they would get if they were in a union. On the other hand, the
more experienced but less educated workers are likely to join trade unions to
secure their jobs.
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
134
The coefficient on specific training (on-the-job training) is negative and statistically
significant; and that for general training is also negative and statistically significant,
a finding that applies to all workers and to production workers. On-the-job training
involves training in specialized tasks that are specific to a particular firm, and it is
most likely that it will reduce the probability of union membership if workers believe
they are likely to lose employment if they join trade unions and hence be unable to
get the specific skills acquired. On the other hand, if workers believe that there is
high job insecurity, so that one can lose a job with or without on-the-job training,
then they are likely to join trade unions. Workers with general training are more
likely to have better job options in other firms and thus their propensity to seek
union-provided goods such as job security, better wages or allowances should be low.
Being a full-time worker raises the probability that a worker is a union member,
especially for production workers. Workers in firms that are highly unionized are
likely to be members of trade unions. This shows that if an individual is employed
in a highly unionized firm, the chances that s/he is a trade union member are
higher than when one is employed in a firm with a low level of unionization. The
family background of a worker also affects union status. Workers whose fathers
have a university degree are likely to be members of trade unions.
The coefficients on location dummies (all workers) indicate that relative to
Mombasa, Eldoret and Nakuru (the location we use for comparison), the
probabilities of being a trade union member are higher in Nairobi. Union
membership probabilities for Nairobi are 4%, higher than the probabilities for
Mombasa, Eldoret and Nakuru. The results for production workers show a
similar pattern to that of all workers, but with much higher elevated
probabilities. It is possible that recruitment of union members is more aggressive
in Nairobi than in the other locations of Mombasa, Eldoret, and Nakuru.
In both sets of results, workers in the food, wood, and textile sub-sector (the
comparison group) are associated with higher probabilities of union membership
relative to employees in the metal sub-sector. For all workers and production
workers, being in the food sub-sector raises the probability of being a trade union
member. The explanation for this outcome could be that the food sub-sector has
well-established and organized trade unions than the metal, textile, and wood
sub-sectors. The strong trade unions in the textile industries in earlier decades
could have been affected by the collapse of major textile firms following the
liberalization of the economy in the 1980s and 1990s.
Weekly hours of work are positively correlated with the probability of being a
trade union member, but this effect is not significant. If an employee is a
production worker, there are higher chances that s/he will be a union member.
Employees in foreign-owned firms are less likely to be union members. If an
employee is in a large firm, such an employee is more likely to be a member of a
trade union. Though the relevant coefficient is not significant, part of the reasons
for higher union membership in large firms is due to lower cost of unionization.
Why Do Workers Join Trade Unions?
Tanzanian Economic Review, Volume 9, Number 2, 2019
135
Longer tenure in a firm raises the probability that a worker is a union member. A
worker in firms where non-union workers benefit from deals reached between
unions and employers is more likely to be a member of a trade union. In Kenya,
workers in the management may not be allowed to join trade unions because of
conflict of interest. However, if it happens that they benefit from deals reached
between unions and employers, then what this result shows is that in such firms,
workers may be encouraged to join trade unions so that whatever benefits they
get also apply to them. Whether an employee is in a firm that uses centralised or
decentralised bargaining processes does not affect union membership. This means
that recruitment practices of unions also affect union membership. Furthermore,
if an employee is married, there is a bigger chance that s/he will be a union
member. This probably reflects the fact that family responsibilities in terms of
food, shelter and school fees pushes a worker to join a trade union for job security,
a key factor in family maintenance.
To examine location effects further, we estimated separate probit equations for
Nairobi and for the other three locations combined. Table 4 shows the estimated
probit results (marginal effects) for the two separate equations.
Table 4: Probit analysis of unionization of workers by location
Variables
Nairobi Other Towns
Marginal
Effects
T-value Marginal
Effects
T-value
Years of education
Experience
Experience squared
On-the-job training
General training
Male
Full-time employment
Union density
Food sector
Wood sector
Textile sector
Weekly hours of work
Production workers
Foreign firm
Firm size
Firm size squared
Tenure
Non-members benefit
Formal sector
Central bargaining
Decentralised bargaining
Married
Number of observations
Log likelihood
Pseudo R-squared
-0.0142
-0.0151
0.0001
-0.0182
-0.1108
-0.0133
0.1578
0.0032
0.0689
-0.0880
-0.0068
-0.0064
0.1908
-0.0729
-0.0001
0.0000
0.0046
0.0995
0.0769
0.3684
0.0968
0.1354
509
-181.54
0.38
-1.97
-2.57
1.31
-0.35
-2.53
-0.23
3.61
5.21
1.21
-1.80
-0.14
-1.59
5.14
-1.43
-0.68
1.11
1.17
2.35
1.21
3.16
1.88
3.14
-0.0012
-0.0007
0.0000
-0.0053
-0.0185
0.0004
-0.0010
0.0004
0.1504
0.0747
0.1197
0.0002
0.0047
0.0570
0.0000
-0.0000
0.0008
0.0281
0.0047
-0.0038
-0.0077
0.0032
432
-70.83
0.63
-2.51
-1.98
0.46
-4.18
-4.30
0.12
-0.27
5.41
3.40
4.52
3.11
1.31
2.26
2.69
3.04
-3.38
3.62
2.78
2.57
-2.13
-1.60
1.35
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
136
The results for Nairobi show that education, experience, and general training
have a negative and significant effect on the probability of being a union member.
Full-time employment, union density, being in firms where non-members benefit
from deals reached by a union, being a production worker, and being a married
person are positively associated with union membership. Similar results are
obtained for the other three urban centres except that the effect of on-the-job
training, which was not significant in the equation for Nairobi, is now significant.
Also, workers in food, wood, and textile industries in the three urban centres are
more likely to be in unions than workers in the metal industry. The other
difference between the two sets of results is that being in a foreign firm increases
the probability of being a union member; and that being a full-time employee has
no effect on union membership for workers in Mombasa, Nakuru, and Eldoret.
Higher weekly hours of work are associated with a higher probability that a
worker would be in a trade union in the other towns, but the association is
statistically insignificant. If a worker is married, then s/he is likely to be a union
member if working in Nairobi. Marital status is not positively associated with
union membership in the other towns. If a worker is in a firm where central
bargaining is done, then such a worker in Nairobi is more likely to be a union
member, while in the other towns s/he is less likely to be in a union. The results
obtained in this study are similar to some of the results obtained by other
previous studies such as Rupayan (2008) for India, Guataqui et al. (2011) for
Colombia, and Windolf and Haas (1989) for West Germany, among others.
6. Conclusion
Union membership in Kenya is voluntary, and about a third of employees are
union workers. The main objective of trade unions is to raise and protect the
welfare of their members through various means, including improving pay and
work conditions, and fringe benefits. It should be the case that positive net
benefits will attract an individual into trade unionism. This paper analysed the
determinants of union membership using data from manufacturing firms in
Kenya for 2000, the most comprehensive information available on the country’s
manufacturing enterprises to date.
The results show that the likelihood of being a union member generally decreases
as the level of a worker’s education increases. The probability of being a trade
union member is higher for full-time employees, married workers, and in firms
where non-members can free-ride. Older employees are more likely to be union
members than young, inexperienced workers. The main motive for a worker’s
membership into a trade union is to protect job tenure and improve working
conditions. The probability of being a union member varies by worker
characteristics and firm attributes; particularly location, type, and size.
Why Do Workers Join Trade Unions?
Tanzanian Economic Review, Volume 9, Number 2, 2019
137
References
Anyango, C., N. Obange, E. Abeka, G. O.Ondiek, O. Odera, & M. E. Ayugi. 2013. Factors
Affecting Performance of Trade Unions in Kenya. American Journal of Business and
Management, 2(2): 181–185.
Booth, A. 1986. Estimating the Probability of Trade Union Membership: A Study of Men
and Women in Britain. Economica, 53: 41–61.
Borland, J. & S. Ouliaris. 1994. The Determinants of Australian Trade Union
Membership. Journal of Applied Econometrics, 9: 453–468.
Borland, J. 1996. Union Effects on Earnings Dispersion in Australia. 1986–1994. British
Journal Industrial Relations, 34: (2): 237–248.
Bulkley B. & G. Myles. 2001. Individually Rational Union Membership. European Journal
of Political Economy, 17(1): 117–137.
Checchi, D. & G. Corneo. 1998. Trade Union Membership: Theories and Evidence for Italy.
Discussion Paper A-526, University of Bonn.
Cole, G. D. H. 1913. The World of Labour. London: G. Bell and Son.
Commons J. R. 1913. Labor and Administration. New York: Macmillan.
Cregan, C. & S. Johnston. 1990. An Industrial Relations Approach to the Free Rider
Problem: Young People and Trade Union Membership in UK. British Journal of
Industrial Relations, 28(1): 85–104.
Deery, S. & H. De Cieri. 1991. Determinants of Trade Union Membership in Australia.
British Journal of Industrial Relations, 29(1): 59–73.
Fitzenberger B., K. Kohn. & Q. Wang. 2009. The Erosion of Union Membership in
Germany: Determinants, Densities and Decompositions. Mimeo. Albert-Ludwigs
University, Germany.
Goerke, L. & M. Pannenberg. 1998. Social Custom, Free-Riders and Trade Union
Membership in Germany and Great Britain. DIW Discussion Paper No. 177, Berlin,
Germany.
Guataqui J. C., A. Garcia. & M. Rodriquez. 2011. Structural Determinants of Trade Union
Membership in Colombia. Perfil de Coyuntura Economica, 17: 31–50.
Hirschman A. O. 1970. Exit, Voice and Loyalty, Responses to Decline in Firms,
Organization and States. Cambridge M. A.: Harvard University Press.
House W. J. & H. Rempel. 1976. The Impact of Unionization on Negotiated Wages in
the Manufacturing Sector in Kenya. Oxford Bulletin of Economics and Statistics,
38(2): 111–123.
Kollmeyer, C. 2007. Who Joins Trade Unions? Testing New Sociological Explanations.
Comparative Sociology, 12(4): 548–574.
Komsi V. M. 2010. Determinants of Trade Union Membership in Finland 1975–2008.
Masters’ thesis, Department of Economics, Alto University.
Lapides, K. (ed.). 1987. Marx and Engels on the Trade Unions. New York: International
Publishers.
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
138
Manda, K. D. 1997. Labour Supply, Returns to Education, and the Effect of Firm Size on
Wages: The Case of Kenya. PhD thesis, University of Gothenburg.
Manda, D. K, A. Bigsten. & G. Mwabu. 2001. The Effect of Trade Unions on Earnings in
Kenyan Manufacturing Firms. Mimeo. Department of Economics, University of
Gothenburg.
Manda, D. K., A. Bigsten. & G. Mwabu. 2005. Trade Union Membership and Earnings in
Kenyan Manufacturing Firms. Applied Economics, 27: (15).
Perlman, S. 1928. A Theory of the Labour Movement. New York: Macmillan.
Riley, N. M. 1997. Determinants of Union Membership: A Review. Labour, 11: 265–301.
Rupayan P. 2008. Estimating the Probability of Trade Union Membership in India: Impact
of Communist Parties, Personal Attribute and Industrial Characteristics. Indira
Gandhi Institute of Development Research, Mumbai, WP=2008–015.
Schmidt P. & R. P. Strauss. 1976. The Effect of Unions on Earnings and Earnings on
Unions: A Mixed Logit Approach. International Economic Review, 17(1): 204–212.
Schnabel, C. & J. Wagner. 2005. Determinants of Trade Union Membership in Western
Germany: Evidence from Micro Data 1980–2000. Socio Economic Review, 3: 1–24.
Schnabel, C. & J. Wagner. 2007. Union Density and Determinants of Union Membership
in 18 EU Countries: Evidence from Micro Data. 2000/03. Industrial Relations Journal,
38(1): 1–24.
Schultz, T.P. & G. Mwabu. 1998. Labour Unions and the Distribution of Wages and
Employment in South Africa. Industrial and Labour Relations Review, 51(4): 680–703.
Smith A. 1776. An Inquiry into the Nature and Causes of the Wealth of Nations.
Edinburgh.
Tennenhaun, F. 1921. The Labour Movement: Its Conservative Functions and Social
Consequences. New York: G P Putman’s Son.
Waddington, J. & C. Whitson.1997. Why Do People Join Unions in a Period of
Membership Decline? British Journal of Industrial Relations, 35: 515–546.
Webb, S. & Webb, B. (eds.). 1897. Industrial Democracy. London, New York, Bombay:
Longmans, Green & Co.
Windolf, P. & J. Haas. 1989. Who Joins the Union? Determinants of Trade Union
Membership in West Germany 1976–1984. European Sociological Review, 5: 147–165.
Why Do Workers Join Trade Unions?
Tanzanian Economic Review, Volume 9, Number 2, 2019
139
Appendix
Table A1: Definition of the variables
Variables Definition
Union dummy
Weekly hours of work
Years of education
Experience
On-the-job training dummy
General training dummy
Production workers
Firm size (workers)
Tenure (years)
Married
Located in Nairobi
Male
Food sector
Wood sector
Textile sector
Metal sector
Union density
Central bargaining
Decentralized bargaining
Fulltime employees
Formal sector
Father has university education
Foreign firm
Non-members benefit
1 if a worker is a union member, 0 otherwise
Total number of hours of work in a week
Total number of years completed in education
Total number of years of experience
1 if a worker has had on-the-job training, 0 otherwise
1 if a worker has had general training, 0 otherwise
1 if one is a production worker, 0 otherwise
Total number of workers in a firm
Total number of years a worker has been with the current firm.
1 if a worker is married. 0 otherwise
1 ff firm is located in Nairobi, 0 otherwise
1 if a worker is married, 0 otherwise
1 if firm is the food sub-sector, 0 otherwise
1 if firm is the wood sub-sector, 0 otherwise
1 if firm is the textile sub-sector, 0 otherwise
1 if firm is the metal sub-sector, 0 otherwise
Proportion of workers in a firm who are union members
1 if for a firm there is central bargaining, 0 otherwise
1 if for a firm bargaining is at the industry of firm level, 0 otherwise
1 if fulltime employee, 0 otherwise
1 if formal sector firm, 0 otherwise
1 if the farther to a worker has university education, 0 otherwise
1 if firm is foreign owned, 0 otherwise
1 if non-union workers benefit from deals reached at between union
and employer organization, 0 otherwise
Damiano Kulundu Manda & Germano Mwabu
Tanzanian Economic Review, Volume 9, Number 2, 2019
140
Table A2: Means of Selected Variables by Union Status for Workers
in the Formal Manufacturing Sector
Variables All Workers Production Workers
Union Non-union Union Non-union
Weekly hours of work
Age
Years of education
Experience
On-the-job training
General training
Production workers
Firm size (workers)
Tenure (years)
Married
Located in Nairobi
Located in Mombasa
Male
Food sector
Wood sector
Textile sector
Metal sector
Union density (%)
Central bargaining
Decentralized bargaining
Fulltime employees
Formal sector
Father has university education
Foreign owned farm
Total Number of observations
46.0
37.5
8.5
15.2
0.14
0.28
0.84
166
12.5
0.91
0.66
0.29
0.91
0.35
0.11
0.38
0.16
66.4
0.09
0.79
0.92
0.92
0.03
0.11
220
46.5
34.1
10.0
11.4
0.11
0.50
0.56
160
7.5
0.71
0.52
0.24
0.79
0.27
0.21
0.23
0.29
24.5
0.05
0.35
0.72
0.92
0.03
0.12
843
46.3
38.0
8.3
16.0
0.15
0.26
-
148
13.2
0.93
0.68
0.27
0.94
0.31
0.12
0.41
0.16
67.4
0.09
0.79
0.93
0.91
0.01
0.07
184
46.9
34.4
9.12
11.9
0.15
0.38
-
127
7.9
0.74
0.48
0.22
0.90
0.22
0.26
0.32
0.29
20.7
0.03
0.32
0.66
0.91
0.01
0.09
486